Henry et al. BMC Public Health (2018) 18:19 DOI 10.1186/s12889-017-4567-2
RESEARCH ARTICLE
Open Access
Area-based socioeconomic factors and Human Papillomavirus (HPV) vaccination among teen boys in the United States Kevin A. Henry1,2* , Allison L. Swiecki-Sikora3, Antoinette M. Stroup4,5, Echo L. Warner6 and Deanna Kepka6,7
Abstract Background: This study is the first to examine associations between several area-based socioeconomic factors and human papillomavirus (HPV) vaccine uptake among boys in the United States (U.S.). Methods: Data from the 2012-2013 National Immunization Survey-Teen restricted-use data were analyzed to examine associations of HPV vaccination initiation (receipt of ≥1 dose) and series completion (receipt of three doses) among boys aged 13-17 years (N = 19,518) with several individual-level and ZIP Code Tabulation Area (ZCTA) census measures. Multivariable logistic regression was used to estimate the odds of HPV vaccination initiation and series completion separately. Results: In 2012-2013 approximately 27.9% (95% CI 26.6%-29.2%) of boys initiated and 10.38% (95% CI 9.48%-11.29%) completed the HPV vaccine series. Area-based poverty was not statistically significantly associated with HPV vaccination initiation. It was, however, associated with series completion, with boys living in high-poverty areas (≥20% of residents living below poverty) having higher odds of completing the series (AOR 1.22, 95% CI 1.01-1.48) than boys in lowpoverty areas (0-4.99%). Interactions between race/ethnicity and ZIP code-level poverty indicated that Hispanic boys living in high-poverty areas had a statistically significantly higher odds of HPV vaccine initiation (AOR 1.43, 95% CI 1.031.97) and series completion (AOR 1.56, 95% CI 1.05-2.32) than Hispanic boys in low-poverty areas. Non-Hispanic Black boys in high poverty areas had higher odds of initiation (AOR 2.23, 95% CI 1.33-3.75) and completion (AOR 2.61, 95% CI 1.06-6.44) than non-Hispanic Black boys in low-poverty areas. Rural/urban residence and population density were also significant factors, with boys from urban or densely populated areas having higher odds of initiation and completion compared to boys living in non-urban, less densely populated areas. Conclusion: Higher HPV vaccination coverage in urban areas and among racial/ethnic minorities in areas with high poverty may be attributable to factors such as vaccine acceptance, health-care practices, and their access to HPV vaccines through the Vaccines for Children Program, which provides free vaccines to uninsured and under-insured children. Given the low HPV vaccination rates among boys in the U.S., these results provide important evidence to inform public health interventions to increase HPV vaccination. Keywords: Human papillomavirus, HPV vaccination, Geographic factors, Cervical cancer, Cancer prevention, Health disparities
* Correspondence: [email protected] 1 Department of Geography, Temple University, 115 W. Polett Walk, 308 Gladfelter Hall, Philadelphia, PA 19122, USA 2 Fox Chase Cancer Center, Cancer Prevention and Control Program, 333 Cottman Avenue, Philadelphia, PA 19111, USA Full list of author information is available at the end of the article © The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Henry et al. BMC Public Health (2018) 18:19
Background In the United States (U.S.), approximately 39,000 HPVassociated cancers among women and men are diagnosed each year, many of which are preventable with the current HPV vaccine. Fifty-nine percent of these cancers occur in women [1]. HPV is the most common sexually transmitted infection in the U.S. [2] and is a known risk factor for genital warts and anal and oropharyngeal cancers in both men and women, cervical cancer in women, and penile cancer in men [1]. Over the past several years rates of HPV-associated cancers among men have been increasing more rapidly than rates of HPV-associated cancers among women [1, 3, 4]. Racial/ethnic and socioeconomic disparities in HPVassociated cancers have been documented in the U.S. Recent data indicate Hispanic men have higher rates of penile cancer compared to non-Hispanic (NH) men. White and Black men have higher rates of oropharyngeal and anal cancers compared to men of other races [1, 5]. Furthermore, incidence rates of HPV-associated cancers overall are highest among men living in census tracts with high poverty levels (>20%)(≥20% of residents living below poverty) compared to men living in census tracts with low poverty levels ($75,000]; above poverty, moderate income [annual income ≤$75,000]; below poverty, based on the U.S. Census family poverty thresholds [42]; and unknown); (e) receipt of a provider recommendation for HPV vaccination (yes, no, or don’t know); (f ) mother’s age (≤34 years, 35-44, or ≥45); (g)mother’s marital status (currently married or not currently married); and (h) mother’s education lev(12 years without college degree; or college degree or higher). We also included one factor from the health-care system level based on the provider survey: facility type (all private facilities, all public facilities, all hospital facilities, all STD/school/teen clinics, or other facilities, plus mixed and unknown) where vaccines were administered. Area-based variables
Several area-based socioeconomic measures created using U.S. Census ZIP Code Tabulation Areas (ZCTAs) data were merged with the ZIP codes of participants’ current residences at the CDC Research Data Center (RDC). ZIP codes are restricted variables; therefore,
Henry et al. BMC Public Health (2018) 18:19
these data were accessed through the RDC. The ZCTAs are generalized areal representations of U.S. Postal Service areas, and in most instances the ZCTA is the same as the ZIP code [43]. ZCTAs were used for this study because smaller geographic areas, such as census tracts, are not available in the NIS-Teen restricted dataset. We included several area-based measures previously incorporated in health disparities research [44, 45] and research on vaccination uptake and use of cancer screening services [16, 46–48]. For this study, ZCTA poverty was conceptualized as an area- or communitybased socioeconomic measure. Area-based socioeconomic measures describe a geographically defined area (e.g., census tracts, ZCTAs) in which an individual lives and that could affect health and access to care through several pathways, including the material resources available (e.g., community health centers), social capital, and social networks (e.g., social contagion, similar norms of behavior) (32, 33, 44). Area-based socioeconomic measures capture conditions that affect all individuals living in the same area and have been shown to be an independent predictor of health outcomes (26-28). SES was based on poverty status derived from the 2008-2012 U.S. Census American Community Survey (ACS). It was grouped into four categories according to the percentage of the population in the ZCTA living below the federally defined poverty threshold: less than 5%, 5-9.9%, 1019.9%, and 20% or greater, with the last category considered a severely disadvantaged area. A measure of racial/ethnic composition or density of a ZCTA was also created. Racial/ethnic composition refers to proportions of people of the same race/ethnicity in a defined area and has been used as a proxy measure of segregation [49, 50]. Segregation refers to the degree to which two or more racial/ethnic groups live separately from one another in a geographically defined area [50]. We used this measure because of our interest in knowing the majority racial/ethnic group in each ZCTA and because area-based socioeconomic measures also vary by racial/ethnic composition in the U. S, with higher socioeconomic areas generally having fewer Blacks and Hispanics and more Whites and Asian/Pacific Islanders (API). Numerous studies have shown that persistent residential racial/ethnic segregation of either poor Blacks or Hispanics in U.S. communities serves as both a health promoter by facilitating stronger social support networks and a health barrier by fostering conditions that limit financial, educational, or social resources [16, 45, 51, 52]. Racial composition was based on the U.S. Census ACS estimates of the percentage of each racial/ethnic group in each ZCTA. Using the percentages, we categorized the majority racial/ethnic group in each ZCTA: >50% Hispanic, >50% NHW, >50% NH Asian/Pacific Islander (NHAPI), >50% NHB, >50% NH American Indian/
Page 4 of 16
Native Alaskan, or NH Mixed if not any of the previous. We combined NH American Indian/Native Alaskan with NH Asian/Pacific Islander into a category called NH Other due to small numbers. We also included a ZCTA measure of population density (total ZCTA population counts divided by land area), which has been used in previous research as an indicator of the built environment, crowding [53] and a proxy of urban/rural residence. ZCTA population density was divided into quartiles based on the nationwide geographic distribution (Q1 1-20, Q2 21-71, Q3 72-651, Q4 > 651 people per square mile) and merged with survey participants based on ZIP code. We also included ZCTA measures of rural and urban residence based on the rural-urban commuting area codes (RUCAs). The RUCAs provided a definition of rural/urban residences based on criteria that included population density and population work-commuting patterns. We categorized this variable as: (1) urban, (2) large rural city/town, (3) small rural town, and (4) isolated small rural town [54]. Statistical analysis
NIS-Teen data for years 2012 and 2013 were combined and the sampling weights for the provider verified data (PROVWT) was recalculated using suggested methods in the NIS-Teen Users Guide [36] to ensure the sampling weights were appropriately adjusted for calculating the weighted percentages and effect estimates. We performed bivariate association tests between all the variables and our two primary HPV vaccination outcomes (initiation and completion) with Wald chi-square tests. We also used logistic regression to identify variables associated with our two primary HPV vaccination outcomes. For each outcome separately, we entered all variables associated with the outcome in bivariate models (p < 0.05) into a multivariable logistic regression model. The multivariate models produced adjusted odds ratios (AORs) and 95% confidence intervals (CIs). One multivariable model included only individual and provider factors, and two separate multivariable models were run to examine the independent associations of ZCTA poverty and racial/ethnic composition with vaccination outcomes. In the two separate multivariable models that included area-based measures we also included population density and in an additional analysis replaced population density with rural/urban residence. Joint contributions of area poverty, racial composition, and race/ethnicity were assessed with interaction terms between individual-level race/ethnicity and ZCTA poverty and individual-level race/ethnicity and ZCTA racial composition. Population density was also included in the multivariable models that were run to examine the associations of ZCTA poverty and racial/ethnic composition with vaccination outcomes. The surveyed state was
Henry et al. BMC Public Health (2018) 18:19
included as a random effect to account for dependency by state residence (e.g., state health programs that will influence the boys from that state). Bivariate associations were analyzed using procedures for complex survey data in SAS statistical software 9.3 (i.e., PROC SURVEYFREQ) [55]. Logistic regression analyses were conducted using SAS GLIMMIX, which implements the generalized linear mixed model and allows for the incorporation of stratum-specific weighted analysis [56]. Statistical tests were two-tailed with a critical alpha of .05.
Results Survey participant characteristics
Sociodemographic and geographic characteristics of the survey participants are presented in Table 1. Overall, the age distribution among boys included in the survey was about even, with each age group making up 19% or 20% of the sample. Most boys were NHW (55.2%), from urban areas (89.0%) with population densities greater than 651 persons per square mile (quartile 4; 59.5%), and in predominately NHW (70.6%) ZCTAs. At the time of the survey, most mothers were >35 years old (90.2%), married (65.5%), did not have a college degree (64.6%), and had employer- or union-provided health insurance (45.9%). Only a third of the parents (35.0%) received a recommendation from their health-care providers to have their sons vaccinated against HPV. Initiation
Overall, for the years 2012-2013, 27.9% of boys received at least one dose of HPV vaccine (Table 1). In the bivariate analysis, all the individual and geographic variables were significantly associated with HPV vaccine initiation except for the boy’s age and facility types for providers (Table 1). Based on multivariable analysis that included only individual-level variables, boys with health insurance through Medicaid/SCHIP or IHS/military insurance had significantly higher odds of HPV vaccine initiation than those with employer- or union-provided insurance (Model 1, AOR 1.53, 95% CI 1.36-1.71; AOR 1.38, 95% CI 1.22-1.56, respectively) (Table 2). Among boys whose parents received a provider recommendation to vaccinate, the odds of HPV vaccination initiation were 9.4 times higher (95% CI 8.66-10.20) than for boys without such a recommendation. Compared to boys whose mothers had a college degree, boys whose mothers had less than 12 years of education had higher odds of vaccine initiation (AOR 1.25, 95% CI 1.09-1.44). However, boys whose mothers were high school graduates or completed some college had significantly lower odds of initiating vaccination (AOR 0.83, 95% CI 0.74-0.93). Boys from households with incomes below the poverty threshold had higher odds of HPV vaccine initiation
Page 5 of 16
compared to households with incomes above the poverty threshold (AOR 1.35, 95% CI 1.17-1.55). Hispanic, NHB, and NH Other had an adjusted odds of initiation of 2.14 (95% CI 1.92-2.39), 1.72 (95% CI 1.53-1.93), and 1.22 (95% CI 1.06-1.40) times greater than NHW boys, respectively. In the multivariable model for initiation (Table 2, Model 3), which included individual-level factors and ZCTA-level poverty and population density, the odds of initiation were highest among boys from the highest poverty category compared to the lowest category, but the result was not statistically significant. The highest population density category (Q4) compared to the lowest population density category (Q1) was not statistically significant; however the highest category (Q4) versus the second (Q2) and third category (Q3) were significant (Model 3, AOR 1.32, 95% CI 1.14-1.52; 1.12 95% 95%CI 1.12 1.02-1.23, respectively). We also separately examined the geographic variable rural/urban residence by including it in Model 3 instead of population density (model results not shown in table). Boys from urban areas had higher odds of initiation when compared to those from isolated small rural towns (reference group) (AOR 1.38, 95% CI 1.04-1.83), small rural towns (reference group) (AOR 1.38, 95% CI 1.09-1.76), and large rural towns (reference group) (AOR 1.50, 95% CI 1.251.80). Figure 1 summarizes the model-adjusted percentage of boys who initiated the HPV vaccination series based on the interaction term race/ethnicity × ZCTA poverty included in the model, and Fig. 2 summarizes the odds ratios for the statistically significant interactions for the same model. Hispanic boys from the most impoverished ZCTAs (≥20% of residents below poverty) had higher odds of HPV vaccination initiation (AOR 1.43, 95%CI 1.03-1.97) than Hispanic boys from the least impoverished ZCTAs (0-4.9% of residents below poverty) (Fig. 2). NHB boys from the most impoverished ZCTAs had higher odds of initiation (AOR 2.23, 95% CI 1.333.75) than NHB boys from the least impoverished ZCATAs. Hispanic boys from the most impoverished ZCTAs also had higher odds of initiation than did NHB (AOR 1.32, 95% CI 1.09-1.59) and NHW (AOR 2.84, 95% CI 2.32-3.46) boys also from the most impoverished ZCTAs. Conversely, NHWs from the most impoverished ZCTAs had lower odds than NHWs from the least impoverished ZCTAs to initiate (AOR 0.80, 95% CI 0.650.97). The interaction term individual-level race/ethnicity × racial composition was not statistically significant. Completion
Overall, for the years 2012-2013, 10.4% of boys completed the recommended HPV vaccine regimen (Table 1). In the bivariate analysis, all the individual and
Henry et al. BMC Public Health (2018) 18:19
Page 6 of 16
Table 1 Individual-level and area-based characteristics of HPV Vaccine Initiation (Receipt of at Least One Dose) and completion (receipt of ≥3 doses): Teen Boys Aged 13 to 17 based on responses from the National Immunization Survey–Teen, 2012-2013 Characteristics
Survey Participants, Weighted % (95% CI), P-value Weighted % (95% CI), Vaccine P-value n weighted % Vaccine Initiation (≥ 1 dose) Completion (≥3 doses)
Total
19,518
10.38 (9.48 – 11.29)
27.9 (26.6 - 29.2)
Year
RESEARCH ARTICLE
Open Access
Area-based socioeconomic factors and Human Papillomavirus (HPV) vaccination among teen boys in the United States Kevin A. Henry1,2* , Allison L. Swiecki-Sikora3, Antoinette M. Stroup4,5, Echo L. Warner6 and Deanna Kepka6,7
Abstract Background: This study is the first to examine associations between several area-based socioeconomic factors and human papillomavirus (HPV) vaccine uptake among boys in the United States (U.S.). Methods: Data from the 2012-2013 National Immunization Survey-Teen restricted-use data were analyzed to examine associations of HPV vaccination initiation (receipt of ≥1 dose) and series completion (receipt of three doses) among boys aged 13-17 years (N = 19,518) with several individual-level and ZIP Code Tabulation Area (ZCTA) census measures. Multivariable logistic regression was used to estimate the odds of HPV vaccination initiation and series completion separately. Results: In 2012-2013 approximately 27.9% (95% CI 26.6%-29.2%) of boys initiated and 10.38% (95% CI 9.48%-11.29%) completed the HPV vaccine series. Area-based poverty was not statistically significantly associated with HPV vaccination initiation. It was, however, associated with series completion, with boys living in high-poverty areas (≥20% of residents living below poverty) having higher odds of completing the series (AOR 1.22, 95% CI 1.01-1.48) than boys in lowpoverty areas (0-4.99%). Interactions between race/ethnicity and ZIP code-level poverty indicated that Hispanic boys living in high-poverty areas had a statistically significantly higher odds of HPV vaccine initiation (AOR 1.43, 95% CI 1.031.97) and series completion (AOR 1.56, 95% CI 1.05-2.32) than Hispanic boys in low-poverty areas. Non-Hispanic Black boys in high poverty areas had higher odds of initiation (AOR 2.23, 95% CI 1.33-3.75) and completion (AOR 2.61, 95% CI 1.06-6.44) than non-Hispanic Black boys in low-poverty areas. Rural/urban residence and population density were also significant factors, with boys from urban or densely populated areas having higher odds of initiation and completion compared to boys living in non-urban, less densely populated areas. Conclusion: Higher HPV vaccination coverage in urban areas and among racial/ethnic minorities in areas with high poverty may be attributable to factors such as vaccine acceptance, health-care practices, and their access to HPV vaccines through the Vaccines for Children Program, which provides free vaccines to uninsured and under-insured children. Given the low HPV vaccination rates among boys in the U.S., these results provide important evidence to inform public health interventions to increase HPV vaccination. Keywords: Human papillomavirus, HPV vaccination, Geographic factors, Cervical cancer, Cancer prevention, Health disparities
* Correspondence: [email protected] 1 Department of Geography, Temple University, 115 W. Polett Walk, 308 Gladfelter Hall, Philadelphia, PA 19122, USA 2 Fox Chase Cancer Center, Cancer Prevention and Control Program, 333 Cottman Avenue, Philadelphia, PA 19111, USA Full list of author information is available at the end of the article © The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Henry et al. BMC Public Health (2018) 18:19
Background In the United States (U.S.), approximately 39,000 HPVassociated cancers among women and men are diagnosed each year, many of which are preventable with the current HPV vaccine. Fifty-nine percent of these cancers occur in women [1]. HPV is the most common sexually transmitted infection in the U.S. [2] and is a known risk factor for genital warts and anal and oropharyngeal cancers in both men and women, cervical cancer in women, and penile cancer in men [1]. Over the past several years rates of HPV-associated cancers among men have been increasing more rapidly than rates of HPV-associated cancers among women [1, 3, 4]. Racial/ethnic and socioeconomic disparities in HPVassociated cancers have been documented in the U.S. Recent data indicate Hispanic men have higher rates of penile cancer compared to non-Hispanic (NH) men. White and Black men have higher rates of oropharyngeal and anal cancers compared to men of other races [1, 5]. Furthermore, incidence rates of HPV-associated cancers overall are highest among men living in census tracts with high poverty levels (>20%)(≥20% of residents living below poverty) compared to men living in census tracts with low poverty levels ($75,000]; above poverty, moderate income [annual income ≤$75,000]; below poverty, based on the U.S. Census family poverty thresholds [42]; and unknown); (e) receipt of a provider recommendation for HPV vaccination (yes, no, or don’t know); (f ) mother’s age (≤34 years, 35-44, or ≥45); (g)mother’s marital status (currently married or not currently married); and (h) mother’s education lev(12 years without college degree; or college degree or higher). We also included one factor from the health-care system level based on the provider survey: facility type (all private facilities, all public facilities, all hospital facilities, all STD/school/teen clinics, or other facilities, plus mixed and unknown) where vaccines were administered. Area-based variables
Several area-based socioeconomic measures created using U.S. Census ZIP Code Tabulation Areas (ZCTAs) data were merged with the ZIP codes of participants’ current residences at the CDC Research Data Center (RDC). ZIP codes are restricted variables; therefore,
Henry et al. BMC Public Health (2018) 18:19
these data were accessed through the RDC. The ZCTAs are generalized areal representations of U.S. Postal Service areas, and in most instances the ZCTA is the same as the ZIP code [43]. ZCTAs were used for this study because smaller geographic areas, such as census tracts, are not available in the NIS-Teen restricted dataset. We included several area-based measures previously incorporated in health disparities research [44, 45] and research on vaccination uptake and use of cancer screening services [16, 46–48]. For this study, ZCTA poverty was conceptualized as an area- or communitybased socioeconomic measure. Area-based socioeconomic measures describe a geographically defined area (e.g., census tracts, ZCTAs) in which an individual lives and that could affect health and access to care through several pathways, including the material resources available (e.g., community health centers), social capital, and social networks (e.g., social contagion, similar norms of behavior) (32, 33, 44). Area-based socioeconomic measures capture conditions that affect all individuals living in the same area and have been shown to be an independent predictor of health outcomes (26-28). SES was based on poverty status derived from the 2008-2012 U.S. Census American Community Survey (ACS). It was grouped into four categories according to the percentage of the population in the ZCTA living below the federally defined poverty threshold: less than 5%, 5-9.9%, 1019.9%, and 20% or greater, with the last category considered a severely disadvantaged area. A measure of racial/ethnic composition or density of a ZCTA was also created. Racial/ethnic composition refers to proportions of people of the same race/ethnicity in a defined area and has been used as a proxy measure of segregation [49, 50]. Segregation refers to the degree to which two or more racial/ethnic groups live separately from one another in a geographically defined area [50]. We used this measure because of our interest in knowing the majority racial/ethnic group in each ZCTA and because area-based socioeconomic measures also vary by racial/ethnic composition in the U. S, with higher socioeconomic areas generally having fewer Blacks and Hispanics and more Whites and Asian/Pacific Islanders (API). Numerous studies have shown that persistent residential racial/ethnic segregation of either poor Blacks or Hispanics in U.S. communities serves as both a health promoter by facilitating stronger social support networks and a health barrier by fostering conditions that limit financial, educational, or social resources [16, 45, 51, 52]. Racial composition was based on the U.S. Census ACS estimates of the percentage of each racial/ethnic group in each ZCTA. Using the percentages, we categorized the majority racial/ethnic group in each ZCTA: >50% Hispanic, >50% NHW, >50% NH Asian/Pacific Islander (NHAPI), >50% NHB, >50% NH American Indian/
Page 4 of 16
Native Alaskan, or NH Mixed if not any of the previous. We combined NH American Indian/Native Alaskan with NH Asian/Pacific Islander into a category called NH Other due to small numbers. We also included a ZCTA measure of population density (total ZCTA population counts divided by land area), which has been used in previous research as an indicator of the built environment, crowding [53] and a proxy of urban/rural residence. ZCTA population density was divided into quartiles based on the nationwide geographic distribution (Q1 1-20, Q2 21-71, Q3 72-651, Q4 > 651 people per square mile) and merged with survey participants based on ZIP code. We also included ZCTA measures of rural and urban residence based on the rural-urban commuting area codes (RUCAs). The RUCAs provided a definition of rural/urban residences based on criteria that included population density and population work-commuting patterns. We categorized this variable as: (1) urban, (2) large rural city/town, (3) small rural town, and (4) isolated small rural town [54]. Statistical analysis
NIS-Teen data for years 2012 and 2013 were combined and the sampling weights for the provider verified data (PROVWT) was recalculated using suggested methods in the NIS-Teen Users Guide [36] to ensure the sampling weights were appropriately adjusted for calculating the weighted percentages and effect estimates. We performed bivariate association tests between all the variables and our two primary HPV vaccination outcomes (initiation and completion) with Wald chi-square tests. We also used logistic regression to identify variables associated with our two primary HPV vaccination outcomes. For each outcome separately, we entered all variables associated with the outcome in bivariate models (p < 0.05) into a multivariable logistic regression model. The multivariate models produced adjusted odds ratios (AORs) and 95% confidence intervals (CIs). One multivariable model included only individual and provider factors, and two separate multivariable models were run to examine the independent associations of ZCTA poverty and racial/ethnic composition with vaccination outcomes. In the two separate multivariable models that included area-based measures we also included population density and in an additional analysis replaced population density with rural/urban residence. Joint contributions of area poverty, racial composition, and race/ethnicity were assessed with interaction terms between individual-level race/ethnicity and ZCTA poverty and individual-level race/ethnicity and ZCTA racial composition. Population density was also included in the multivariable models that were run to examine the associations of ZCTA poverty and racial/ethnic composition with vaccination outcomes. The surveyed state was
Henry et al. BMC Public Health (2018) 18:19
included as a random effect to account for dependency by state residence (e.g., state health programs that will influence the boys from that state). Bivariate associations were analyzed using procedures for complex survey data in SAS statistical software 9.3 (i.e., PROC SURVEYFREQ) [55]. Logistic regression analyses were conducted using SAS GLIMMIX, which implements the generalized linear mixed model and allows for the incorporation of stratum-specific weighted analysis [56]. Statistical tests were two-tailed with a critical alpha of .05.
Results Survey participant characteristics
Sociodemographic and geographic characteristics of the survey participants are presented in Table 1. Overall, the age distribution among boys included in the survey was about even, with each age group making up 19% or 20% of the sample. Most boys were NHW (55.2%), from urban areas (89.0%) with population densities greater than 651 persons per square mile (quartile 4; 59.5%), and in predominately NHW (70.6%) ZCTAs. At the time of the survey, most mothers were >35 years old (90.2%), married (65.5%), did not have a college degree (64.6%), and had employer- or union-provided health insurance (45.9%). Only a third of the parents (35.0%) received a recommendation from their health-care providers to have their sons vaccinated against HPV. Initiation
Overall, for the years 2012-2013, 27.9% of boys received at least one dose of HPV vaccine (Table 1). In the bivariate analysis, all the individual and geographic variables were significantly associated with HPV vaccine initiation except for the boy’s age and facility types for providers (Table 1). Based on multivariable analysis that included only individual-level variables, boys with health insurance through Medicaid/SCHIP or IHS/military insurance had significantly higher odds of HPV vaccine initiation than those with employer- or union-provided insurance (Model 1, AOR 1.53, 95% CI 1.36-1.71; AOR 1.38, 95% CI 1.22-1.56, respectively) (Table 2). Among boys whose parents received a provider recommendation to vaccinate, the odds of HPV vaccination initiation were 9.4 times higher (95% CI 8.66-10.20) than for boys without such a recommendation. Compared to boys whose mothers had a college degree, boys whose mothers had less than 12 years of education had higher odds of vaccine initiation (AOR 1.25, 95% CI 1.09-1.44). However, boys whose mothers were high school graduates or completed some college had significantly lower odds of initiating vaccination (AOR 0.83, 95% CI 0.74-0.93). Boys from households with incomes below the poverty threshold had higher odds of HPV vaccine initiation
Page 5 of 16
compared to households with incomes above the poverty threshold (AOR 1.35, 95% CI 1.17-1.55). Hispanic, NHB, and NH Other had an adjusted odds of initiation of 2.14 (95% CI 1.92-2.39), 1.72 (95% CI 1.53-1.93), and 1.22 (95% CI 1.06-1.40) times greater than NHW boys, respectively. In the multivariable model for initiation (Table 2, Model 3), which included individual-level factors and ZCTA-level poverty and population density, the odds of initiation were highest among boys from the highest poverty category compared to the lowest category, but the result was not statistically significant. The highest population density category (Q4) compared to the lowest population density category (Q1) was not statistically significant; however the highest category (Q4) versus the second (Q2) and third category (Q3) were significant (Model 3, AOR 1.32, 95% CI 1.14-1.52; 1.12 95% 95%CI 1.12 1.02-1.23, respectively). We also separately examined the geographic variable rural/urban residence by including it in Model 3 instead of population density (model results not shown in table). Boys from urban areas had higher odds of initiation when compared to those from isolated small rural towns (reference group) (AOR 1.38, 95% CI 1.04-1.83), small rural towns (reference group) (AOR 1.38, 95% CI 1.09-1.76), and large rural towns (reference group) (AOR 1.50, 95% CI 1.251.80). Figure 1 summarizes the model-adjusted percentage of boys who initiated the HPV vaccination series based on the interaction term race/ethnicity × ZCTA poverty included in the model, and Fig. 2 summarizes the odds ratios for the statistically significant interactions for the same model. Hispanic boys from the most impoverished ZCTAs (≥20% of residents below poverty) had higher odds of HPV vaccination initiation (AOR 1.43, 95%CI 1.03-1.97) than Hispanic boys from the least impoverished ZCTAs (0-4.9% of residents below poverty) (Fig. 2). NHB boys from the most impoverished ZCTAs had higher odds of initiation (AOR 2.23, 95% CI 1.333.75) than NHB boys from the least impoverished ZCATAs. Hispanic boys from the most impoverished ZCTAs also had higher odds of initiation than did NHB (AOR 1.32, 95% CI 1.09-1.59) and NHW (AOR 2.84, 95% CI 2.32-3.46) boys also from the most impoverished ZCTAs. Conversely, NHWs from the most impoverished ZCTAs had lower odds than NHWs from the least impoverished ZCTAs to initiate (AOR 0.80, 95% CI 0.650.97). The interaction term individual-level race/ethnicity × racial composition was not statistically significant. Completion
Overall, for the years 2012-2013, 10.4% of boys completed the recommended HPV vaccine regimen (Table 1). In the bivariate analysis, all the individual and
Henry et al. BMC Public Health (2018) 18:19
Page 6 of 16
Table 1 Individual-level and area-based characteristics of HPV Vaccine Initiation (Receipt of at Least One Dose) and completion (receipt of ≥3 doses): Teen Boys Aged 13 to 17 based on responses from the National Immunization Survey–Teen, 2012-2013 Characteristics
Survey Participants, Weighted % (95% CI), P-value Weighted % (95% CI), Vaccine P-value n weighted % Vaccine Initiation (≥ 1 dose) Completion (≥3 doses)
Total
19,518
10.38 (9.48 – 11.29)
27.9 (26.6 - 29.2)
Year
